Python implementation of Structure-based visual localization using SIFT matches. A sparse 3D reconstruction, from monocular image data is initially created. 2D-3D matches are established from new query images. Finally the camera pose is estimated using PnP inside a RANSAC loop.